Welcome to the Deep Learning Labs at the RSNA 2026 Annual Meeting (Radiological Society of North America) in Chicago!
The Deep Learning Labs are one-hour, hands-on classes where you work with AI tools directly, from building AI agents and prototyping applications to evaluating, fine-tuning, and preparing data for deep learning models. This repository holds the materials for every lab, including Google Colab notebooks, slides, and supplementary files. Each lab has its own numbered folder, in the order the labs are held.
Dates and times may change. For the most up-to-date schedule, please check the official RSNA 2026 Annual Meeting program. All times are U.S. Central Time.
| # | Date | Time | Lab | Level | Moderator | Presenters |
|---|---|---|---|---|---|---|
| 01 | Sun, Nov 29 | 10:30 – 11:30 AM | A Low-Code Introduction to AI Agents in Radiology | Core | Bardia Khosravi, MD MPH MHPE | Vera Sorin, MD; Amirali Khosravi, MD |
| 02 | Sun, Nov 29 | 2:30 – 3:30 PM | Building Production-Ready Agentic AI Applications: A Review of the State of the Art in Radiology | Advanced | Pouria Rouzrokh, MD MPH MHPE | Melina Hosseiny, MD |
| 03 | Mon, Nov 30 | 8:00 – 9:00 AM | An Introduction to Foundation Models in Medical Imaging | Core | Judy Gichoya, MD FSIIM | Mohammadreza Chavoshi, MD; Frank Li, PhD |
| 04 | Mon, Nov 30 | 9:30 – 10:30 AM | How to Evaluate a Deep Learning Model for Deployment in Real Practice | Core | Shahriar Faghani, MD | Mana Moassefi, MD |
| 05 | Tue, Dec 1 | 1:30 – 2:30 PM | Vibe Coding 101: Building Radiology Applications Without Programming Experience | Core | Tugba Akinci D'Antonoli, MD | Keno Bressem; Walter Wiggins, MD PhD |
| 06 | Tue, Dec 1 | 4:30 – 5:30 PM | Leveraging Pretrained Embedding Models for Your Data and Use Cases | Advanced | Jason Klotzer | Kenneth Philbrick, PhD |
| 07 | Wed, Dec 2 | 8:00 – 9:00 AM | How to Fine Tune a Vision-Language Model | Advanced | Ali Ganjizadeh, MD MBA | Kalina Slavkova, PhD; Julie Bauml, MD |
| 08 | Wed, Dec 2 | 9:30 – 10:30 AM | An Introduction to Sourcing and Preprocessing Medical Imaging Data for Vision-Language Models | Core | Felipe Kitamura, MD PhD | Imon Banerjee; Eduardo Farina, MD; Alessia Guarnera |
- A Low-Code Introduction to AI Agents in Radiology: What makes an AI agent different from a chatbot, and how to build your own agents with no-code and low-code (Pydantic AI) tools.
- Building Production-Ready Agentic AI Applications: Multi-agent AI frameworks and the practical know-how needed to deploy connected LLM agents in real-world scenarios.
- An Introduction to Foundation Models in Medical Imaging: What foundation models are, how they work, and how to load and apply one to defined tasks. No prior knowledge assumed.
- How to Evaluate a Deep Learning Model for Deployment in Real Practice: Evaluating models beyond research metrics, covering data drift, calibration, uncertainty, and human-in-the-loop monitoring.
- Vibe Coding 101: Prompt engineering and "vibe coding" to prototype radiology applications without prior programming experience.
- Leveraging Pretrained Embedding Models for Your Data and Use Cases: Adapting open-weight, radiology-specific pretrained models to your own use case, with attention to data curation, validation, and operating points.
- How to Fine Tune a Vision-Language Model: An end-to-end walkthrough of fine-tuning a vision-language model on a radiology dataset, from data preparation to failure analysis.
- Sourcing and Preprocessing Medical Imaging Data for Vision-Language Models: Building image-report datasets from the ground up, covering cohort selection, DICOM retrieval with pynetdicom, de-identification, and quality control.
All lessons are designed to run in Google Colab, a free, web-based Jupyter notebook environment hosted by Google. To take part:
- Bring a laptop. We recommend a recent computer running the Chrome browser.
- Have a Google account. You need one (e.g., Gmail) to use Colab. If you don't have one, please create an account before the meeting. You can delete the account afterward if you wish.
- Check your lab's folder before the session. Some labs may ask you to complete extra setup in advance, such as creating a free account or API key. Any such steps will be listed in the lab's README.
Notebooks will be linked from each lab's folder with an Open in Colab button once materials are posted.
Please read CONTRIBUTING.md for instructions on uploading your materials, notebook guidelines, and key deadlines.
This repository is released under the MIT License. Individual lab materials may reference third-party datasets, models, or software with their own licenses. Please review them before reuse.